Improving the Traditional Propagation Model on the Multi-layer Network: From Random Initialization to Relationship-Driven Influence

Node influence propagation model plays a significant role in fields such as key node identification. Multi-layer network can accurately capture the multidimensional dependencies in complex systems and reveal structural features and dynamic behaviors. This paper proposes a novel node influence propagation framework named IED2 that based on relationship-driven to improve the application of existing propagation models in multi-layer network. IED2 effectively solves the limitations of traditional propagation models in random initialization of node influence in multi-layer network. In addition, this paper proposes a new propagation effect evaluation criterion to comprehensively measure the propagation ability and speed of network. The experimental results show that the model combined with IED2 significantly improves in efficiency of the propagation.